Metabolic Profiling of Human Urine - American Chemical Society

Subscriber access provided by NORTH CAROLINA STATE UNIV

Article

Following healthy pregnancy by Nuclear Magnetic Resonance (NMR) metabolic profiling of human urine Sílvia O. Diaz, António S. Barros, Brian J. Goodfellow, Iola Fernandes Duarte, Isabel M. Carreira, Eulália Galhano, Cristina Pita, Maria do Céu Almeida, and Ana Maria Gil J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/pr301022e • Publication Date (Web): 11 Dec 2012 Downloaded from http://pubs.acs.org on December 22, 2012

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Proteome Research is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Following healthy pregnancy by Nuclear Magnetic Resonance (NMR) metabolic profiling of human urine

Sílvia O. Diaz1, António S. Barros2, Brian J.Goodfellow1, Iola F. Duarte1, Isabel M. Carreira3,4, Eulália Galhano5, Cristina Pita5, Maria do Céu Almeida5, Ana M. Gil1,*

1

CICECO-Department of Chemistry, Campus Universitário de Santiago, Universidade de

Aveiro, 3810-193 Aveiro, Portugal 2

QOPNA- Department of Chemistry, Campus Universitário de Santiago, University of Aveiro,

3810-193 Aveiro, Portugal 3

Cytogenetics and Genomics Laboratory, Faculty of Medicine, University of Coimbra, Portugal

and CENCIFOR - Forensic Science Centre, Portugal 4

CIMAGO - Centro de Investigação Meio Ambiente, Genética e Oncobiologia, Portugal

5

Maternity Bissaya Barreto, Centro Hospitalar de Coimbra, 3000-061 Coimbra, Portugal

*Corresponding author: tel +351 234 370707, fax +351 234 370084, e-mail [email protected]

1 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 36

Abstract In this work, untargeted NMR metabonomics was employed to evaluate the effects of pregnancy on the metabolite composition of maternal urine, thus establishing a control excretory trajectory for healthy pregnancies. Urine was collected for independent groups of healthy non-pregnant and pregnant women (in 1st, 2nd, 3rd trimesters) and multivariate analysis performed on the corresponding NMR spectra. Models were validated through Monte Carlo Cross Validation and permutation tests and metabolite correlations measured through Statistical Total Correlation Spectroscopy. The levels of 21 metabolites were found to change significantly throughout pregnancy, with variations observed for the first time to our knowledge for choline, creatinine, 4deoxyerythronic acid, 4-deoxythreonic acid, furoylglycine, guanidoacetate, 3-hydroxybutyrate and lactate. Results confirmed increased aminoaciduria across pregnancy and suggested a) a particular involvement of isoleucine and threonine in lipid oxidation/ketone body synthesis, b) a relation of excreted choline, taurine and guanidoacetate to methionine metabolism and urea cycle regulation and c) a possible relationship of furoylglycine and creatinine to pregnancy, based on a tandem study of non-fasting confounding effects. Results demonstrate the usefulness of untargeted metabonomics in finding biomarker metabolic signatures for healthy pregnancies, against which disease-related deviations may be confronted in future studies, as a base for improved diagnostics and prediction.

Keywords: urine, pregnancy, prenatal health, metabonomics, metabolomics, NMR, multivariate analysis, metabolic trajectory


Page 3 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Introduction Pregnancy is an extraordinarily dynamic period during which women undergo major anatomic, physiologic and metabolic adaptations, in order to ensure the continuous availability of conditions and substrates required for fetal development1. Clinical pregnancy monitoring relies on ultrasound testing, often in combination with chorionic villus sampling, amniocentesis and cordocentesis for fetal karyotyping and/or other genetic disorders, as well as measurement of specific maternal serum markers associated with increased risk of chromosomal disorders e.g. βhuman chorionic gonadotropin (hCGβ) and pregnancy-associated plasma protein-A (PAPP-A)2. The possibility of exploiting maternal/fetal metabolic characteristics to define new biomarkers and develop new methods for non-invasive pregnancy monitoring, aiming at early disease detection and prediction has been recognized in recent years and a steady inflow of research developments has been reported in this context3-9. Nuclear Magnetic Resonance (NMR) and Mass Spectrometry (MS) metabonomics (or metabolomics) of biological samples have been increasingly used to probe for statistically relevant metabolic changes accompanying the onset of disease10,11. In a prenatal health context, such approaches have been applied to amniotic fluid, maternal blood plasma and urine, umbilical cord blood and placental cell cultures, usually collected at a single point in time and in connection to relevant prenatal diseases such as preeclampsia7,8,12-15, fetal malformations4,5,15-18, chromosomal disorders5,18, preterm delivery35,9,18

, gestational diabetes mellitus4,5,18, small for gestational age6,9,19. However, the observed

metabolite changes in the case of disease onset and progression necessarily overlap with those characterizing the expected progression of pregnancy so that gestational age is a possible confounding variable to take into account. Furthermore, the search for prenatal disease biomarkers should benefit from multiple-timepoint collection, rather than single-timepoint as



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 36

mostly done in the past, so that more reliable models may be built for disease diagnosis or prediction. Therefore, a thorough untargeted time course characterization of the effects of healthy pregnancies (controls) on the metabolite compositions of maternal (e.g. urine, blood) and fetal (e.g. amniotic fluid, umbilical cord blood) biofluids is justified. Such a study should provide evidence for several well documented metabolic aspects of healthy pregnancies e.g. fat accumulation and lipogenesis in early pregnancy followed by accelerated fat breakdown and glycerol use in maternal gluconeogenesis for glucose redirection to the fetus, or the marked changes in protein metabolism, favoring nitrogen conservation and active transfer of amino acids to the fetus1,20,21. In addition, however, the unveiling of novel metabolic information and biomarkers for healthy pregnancy may be achieved. Past reports, in the context of the metabolic follow-up of pregnancy, comprise mostly targeted studies of biofluid composition, having shown that 3rd trimester (3rd T) amniotic fluid contains decreased contents of alanine, citrate and glucose and increased histidine, tyrosine and valine22, and that the levels of alanine, glutamine, valine, creatinine, glucose and succinic acid in that biofluid reflect gestational age23-25. Regarding plasma, decreases in several amino acids have been noted (hypoaminoacidemia)26, in tandem with increased plasma lipids and decreased circulating carnitines (free carnitine and acylcarnitine). In relation to urine, increased excretion of carnitines has been seen in the 1st T, followed by a decrease thereafter27, whereas other targeted studies have detected increased excretion of several amino acids, glucose and folate28,29. However, no overall metabolite profiling of maternal urine has been performed, to our knowledge, as a function of pregnancy progression. This work describes, for the first time to our knowledge, an untargeted 1H NMR study of urine collected for healthy non-pregnant and 1st, 2nd and 3rd trimester pregnant women, to evaluate the


Page 5 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


dynamic metabolic adaptations of healthy pregnancies. The results should provide confirmation of known metabolite excretory characteristics, as well as the identification of additional changes. This work should enable an overall typical metabolic trajectory to be defined for healthy pregnancies, thus setting the basis for the detection of deviant trajectories related to pathological situations, in a clinical context. As part of the study, the possible confounding effects of nonfasting conditions for sample collection were also evaluated and discussed.

Experimental Section Samples Urine samples were collected for healthy pregnant women, at 1st, 2nd and 3rd trimesters, during their routine medical appointments. Table 1 lists the number of samples for each independent subject group, along with corresponding metadata information. Independent groups were chosen, rather than multiple-collection for each subject, in order to consider pregnancy-related changes larger than inter-subject variability and, therefore, reflecting the general population more closely. Samples were collected under the approval of the ethical committee of the Hospital Center of Coimbra (Refs.18/04 and 29/09) and informed consents were obtained from each subject participating in the study. Due to medical restrictions in controlling/restricting pregnant women’s diet, all urine samples from pregnant women were collected non-fasting. Pregnancies were followed up until term to confirm that women had normal term pregnancies and healthy babies. Clinical metadata was obtained from obstetrical and neonatal medical records and from an individual questionnaire filled in at the time of collection. For a group of NP women, urine was collected fasting and non-fasting (2 hours after random breakfast), in order to evaluate the effect of the typical non-fasting conditions in which sample collection took place for pregnant women.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 36

Urine samples (50 mL) were stored at -20ºC for up to 2 h and then transferred to -80ºC. Before analysis, samples were thawed and 800 µL were centrifuged (4500g, 5 min). Then, 60 µL of 1.5 M KH2PO4/D2O buffer pH 7, 0.1%Na+/3-trimethylsilyl-propionate (TSP)5 were added to 540 µL of supernatant, followed by pH readjustment to 7±0.02 with KOD (4 M) or DCl (4 M). The mixture was again centrifuged (4500g, 5 min) and 550 µL were transferred to a 5 mm NMR tube. NMR Spectroscopy NMR spectra were recorded on a Bruker Avance DRX 500 spectrometer at 300 K. Standard 1D spectra were acquired, using a noesy 1D pulse sequence with tm 100 ms, a fixed 3 µs t1 delay, and water suppression during relaxation delay and mixing time. 128 transients were acquired into 64k complex data points, with a 10080.65 Hz ppm spectral width, a 4s relaxation delay and a 3.25 s acquisition time. Each FID was zero-filled to 64k points, multiplied by a 0.3 Hz exponential line-broadening function prior to Fourier transform. Spectra were manually phased and baseline corrected and chemical shifts referenced internally to TSP at δ=0.0 ppm. Peak assignments were carried out with basis on literature, 2D NMR experiments (namely, Total Correlation Spectroscopy-TOCSY, Heteronuclear Single Quantum Coherence- HSQC and Jresolved), consultation of spectral databases (Bruker Biorefcode database and the human metabolome database (HMDB)30, spectra of standard compounds and spiking experiments. Analysis of NMR data Multivariate analysis was applied to the full resolution 1H NMR urine spectra with exclusion of the water (4.60-5.05 ppm) and urea (5.50-6.20 ppm) regions. Spectra were aligned using a recursive segment-wise peak alignment31, to minimize chemical shift variations, and data was normalized through probabilistic quotient normalization (PQN), to account for sample concentration differences32. Principal component analysis (PCA)33 and partial-least squares


Page 7 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


discriminant analysis (PLS-DA)34 were performed after unit variance (UV) scaling of the spectra, using SIMCA-P 11.5. The corresponding loading weights were obtained by multiplying each variable by its standard deviation and were colored according to each variable importance to the projection (VIP). PLS-DA model validation was carried out by Monte-Carlo crossvalidation (MCCV) (7 blocks) using 500 runs, with recovery of Q2 values and confusion matrices. Classification rates, specificity and sensitivity were computed and the predictive power of each model was further assessed using a receiver operating characteristic (ROC) map, a function of the true positive rate (TPR or sensitivity) and false positive rate (FPR or 1specificity). PLS-DA models were considered robust when minimal overlapping of the distribution of true and permuted Q2 was obtained35,36. For validated models, the relevant peaks were integrated from the original spectra using Amix 3.9.5, BrukerBioSpin, Rheinstetten, Germany, and normalized to total intensity. All integrals were compared through the Wilcoxon test. For each set of significant metabolites (p < 0.05), the size effect (considered in order to take into account sample dispersion) was determined as: size effect =

x1 − x 2 , where x1 and x 2 are s

the average metabolite integrals for sample groups 1 and 2 respectively, s is the pooled variance: s2 =

( n1 − 1) × s12 + ( n2 − 1) × s22 with group sample numbers n1 and n2, and the corresponding n1 + n2 − 2

standard deviations s12 and s 22 37. Statistical tests, boxplots, heatmaps and loadings plots (VIP coloured) were carried out using R-statistical software, along with MATLAB (version 7.12.0, The MathWorks, Inc.), Lattice and Plotrix (freeware) packages. Statistical total correlation spectroscopy (STOCSY)38 calculation was performed using MATLAB, to aid assignment and identification of metabolically related compounds. A correlation coefficient (r) cut-off of |r| ≥ 0.6 (p < 0.05) was applied. 7 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 36

Results Figure 1 shows typical 1H NMR spectra of urine collected for non-pregnant (NP) women and pregnant women in their 1st, 2nd and 3rd trimesters of pregnancy. The assignment of 57 different metabolites was achieved (Supporting information, Table S2) and the rectangles in Figure 1 indicate some of the spectral changes visible throughout pregnancy. In order to assess the statistically relevant changes in the 1H NMR spectra, as a function of time, multivariate analysis was carried out considering each pair of sample groups. Figure 2a (left) shows the PLS-DA scores plot relating to NP/1stT samples, showing clear group separation, with a high Q2 (0.73) expressing the reliability of group separation. MCCV analysis calculated 90% sensitivity and specificity (Supporting information, Table S1), along with low-overlapping Q2 distribution for true and permuted models (Supporting information, Figure S1a), reflecting the good predictive power of the NP/1stT model. Interpretation of the corresponding loading weights (Figure 2a, right) enabled the identification of the main compounds varying between the two states and the variations in the corresponding integrals have been tested for significance (p-value < 0.05) and size effect (Table 2). Figure 2b,c shows the PLS-DA results obtained for 1st/2ndT and 2nd/3rdT, both models showing clear separation supported by Q2 values of 0.48 and 0.58, respectively (Supporting information, Table S1). The 1st T and the 3rd T outliers marked in Figure 2 were revealed to contain, respectively, higher hippurate and trimethylamine-N-oxide (TMAO) contents, both changes possibly relating to diet39. MCCV of the 1st/2ndT model showed lower sensitivity, 75%, and especially lower specificity, 64% (higher rate of false positives), along with a high-overlapping distribution of Q2 values (not shown). In spite of known metabolite changes taking place early on in pregnancy, for instance when considering placenta composition8, the


Page 9 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


lower predictive power of this model, in the particular conditions of this study, is not surprising when considering the closeness of gestational ages (medians of 12 and 17 g.w., respectively for 1st and 2nd trimesters, Table 1). On the other hand, MCCV of the 2nd/3rdT model showed improved specificity and sensitivity (82% and 86%, respectively), with 100% of permuted models having lower Q2 than the original model (not shown). The analysis of the loading plots corresponding to the results shown in Figure 2 revealed that a large number of resonances change, mainly at the beginning and at the end of pregnancy (Table 2, box plots in additional file 1, Figure S2). Relevant metabolite variations comprised, in the 1st T compared to NP, increases in unassigned compounds U1 to U4, alanine, glycine, histidine, tyrosine, as well as 4-deoxyerythronic acid (4-DEA), choline and guanidoacetate (GAA). In tandem decreases were noted for creatinine and furoylglycine. Of these, U1, U2, alanine and 4DEA increased steadily throughout pregnancy, whereas the remaining metabolites tended to stabilize. 2nd trimester urine also showed an increase in 4-deoxythreonic acid (4-DTA, a diastereomer of 4-DEA) and decreases in U3 and taurine. 3rd trimester samples were, however, characterized by marked additional changes: increases in isoleucine, leucine, threonine, 3hydroxybutyrate (3-HBA), lactate and furoylglycine, in tandem with decreases in carnitine. Notably, the changes listed in Table 2 include some detected for the first time to our knowledge in maternal urine, in a pregnancy context, namely those regarding: choline, creatinine, 4-DEA and 4-DTA, furoylglycine, GAA, 3-HBA and lactate (noted as * in Table 2). Furthermore, an increase was found here for leucine, contrary to other reports indicating decreased excretion late in pregnancy28,29; however, our observation is actually in agreement with an earlier report40. In addition, the four still unassigned resonances (U1 to U4) play an important role in defining the compositional profile of maternal urine throughout pregnancy, particularly U1 and U2, which



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 36

have significant variation between NP and 1st T (p-value ≈ 10-6), and between 2nd and 3rd T (pvalue ≈ 10-7). The identification of additional resonances belonging to U1 and U2 spin systems was attempted by 2D NMR and STOCSY analysis (Table 2), however, no further compound assignments were achieved. Overall, a total of 21 compounds and unassigned resonances were seen to vary significantly at some stage during pregnancy. These variations may be visualized through a qualitative plot of the corresponding normalized integrals, throughout pregnancy (Figure 3). In spite of the visible inter-subject variability, the significant variations mentioned earlier can be clearly identified e.g. increases for U1 and U2, some aminoacids (e.g. alanine, glycine, histidine, tyrosine) and other compounds (e.g. 3-HBA, 4-DEA, choline), in tandem with decreases in taurine, carnitine, furoylglycine and creatinine. Multivariate analysis has also been applied to the selection of 21 metabolites/resonances and resulted in the significant improvement of the two-class models (reaching 96-100% sensitivity and specificity), with the exception of the 1st vs 2ndT model, which retained low specificity (68%) and Q2 values (Supporting information, Table S1), due to the great similarity between the 1st and 2nd trimester groups defined. The consideration of the selected 21 integrals, instead of all spectral points, also resulted in an improvement in the overall excretory trajectory of pregnancy, as visualized through PCA or PLS-DA (Figure 4c,d, compared to Figure 4a,b). The curved arrow in Figure 4d indicates the trajectory followed throughout pregnancy and the sample indicated with a short arrow corresponds to one collected relatively late in pregnancy (39 g.w.), thus seemingly continuing the samples trend towards pregnancy term. This sample was shown to have high contents of choline and U1, these components revealing to be highly determinant at the end of pregnancy. Based on the results in Figure 4d, each subject group could then be described by a discrimination function of the type y = c + b1x1 + ... + bn xn , with n=21, where each coefficient bi


Page 11 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


gives the contribution of variable i for what can be considered a profile signature for each trimester of a healthy pregnancy (Supporting information, Figure S3). These functions describe the weight-distribution of the 21-metabolite integrals characteristic of NP and each pregnancy stage, possibly useful in future profile-matching of the urine of new subjects (in a given trimester) to confirm the healthy course of pregnancy or detect deviant trajectories. Furthermore, the possible confounding nature of non-fasting conditions on the trajectory measured was investigated through PLS-DA of the urine spectra of the NP group, collected under both fasting and non-fasting conditions (see Experimental section). This produced weak group separation (Supporting information, Figure S4a), with Q2 0.30 and considerable overlap of Q2 distributions (not shown), with creatinine and furoylglycine showing up as the only metabolites common to those observed to change in healthy pregnancies (Supporting information, Figure S4b). In the case of furoylglycine, however, its variation throughout pregnancy (Table 2) seems quite distinct from the ca. 2-fold increase observed upon breakfast ingestion (Supporting information, Figure S4b), as will be discussed below. As expected, sample dispersion is seen to decrease to some extent under fasting conditions (Supporting information, Figure S4a), so that models’ quality might benefit slightly from such conditions (although possibly compromising their practicality in the clinic, in the present context). Further metabolic characterization of each subject group was sought through STOCSY, in order to identify eventual metabolically related compounds. In this way, meaningful correlations were found for the a) 1st trimester: U1/U2 (r 0.70); Isoleucine/Leucine (r 0.68); Isoleucine/U4 (r 0.73) and Isoleucine /carnitine (r 0.72); 3-HBA/U4 (r 0.93) and 3-HBA/Carnitine (r 0.74); Alanine/GAA (r 0.61); 4-DTA/carnitine (r 0.74); b) 2nd trimester: U1/U2 (r 0.88);



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 36

Isoleucine/Leucine (r 0.63); and c) 3rd trimester: U1/U2 (r 0.76); U2/3-HBA (r 0.74) and 4DEA/3-HBA (r 0.73); Isoleucine/Leucine (r 0.85).

Discussion The significant changes observed in maternal urine composition throughout pregnancy relate mainly to i) amino acids and some of their derivatives, ii) choline, iii) carnitine and 3-HBA, iv) creatinine and v) four important unassigned resonances (Table 2). Although some of these changes were registered here, in connection to pregnancy, for the 1st time to our knowledge, (namely, for choline, creatinine, 4-DEA and 4-DTA, furoylglycine, GAA, 3-HBA and lactate), most observations fit broadly with the existing knowledge of the biochemical phenomena accompanying pregnancy1,20,21,29, as discussed below.

Aminoacid and Energy Metabolisms Pregnancy is usually accompanied by selective aminoaciduria, the exact mechanism of which is not fully known. In this context, previous reports28,29 have indicated that a) excretion of alanine, glycine, histidine, serine and threonine increases during pregnancy; b) excretion of cysteine, leucine, lysine, phenylalanine, taurine, tyrosine and valine increases early on and decreases later in gestation and c) excretion of arginine, asparagine, glutamate, isoleucine, methionine and ornithine is not expected to change significantly. In this work, results have shown that a) alanine, glycine, histidine and threonine were confirmed to increase in urine; b) changes in excreted taurine and tyrosine were confirmed, while leucine varied differently than expected (↑ in 3rd T); and c) out of the amino acids not expected to change significantly, isoleucine presented a marked increase in the 3rd trimester, consistently however with an earlier report40. In relation to


Page 13 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


the remaining amino acids, arginine, asoaragine, cysteine, glutamate, methionine, phenylalanine and ornithine were not detected (probably due to their too low concentrations for NMR observation: up to tens of µmol/L), whereas no significant changes could be measured for lysine, serine and valine. Pregnancy is known to be accompanied by increased placental transfer of amino acids, favoring nitrogen conservation for fetal growth1,20, and thus leading to diminished circulating levels of many amino acids in plasma, a condition known as hypoaminoacidemia. This should also relate to an underuse of many amino acids in maternal gluconeogenesis, particularly later in gestation, making way for alternative substrates such as glycerol1. Furthermore, the lower plasma levels of alanine and other urea cycle amino acids (citrulline, ornithine and arginine) throughout pregnancy have also been associated to a slowing down of urea synthesis20. In spite of the expected redirection of alanine and other amino acids to the fetus, their excretion is typically increased, an aspect which has been proposed to reflect impaired renal reabsorption rather than a saturation phenomenon28,29. Regarding glycine, its involvement in methionine metabolism might suggest an enhanced production from dimethylglycine, DMG (formed through remethylation of homocysteine to methionine)26. However, this is an unlikely source of glycine in early pregnancy, since the alternative pathway of homocysteine conversion to cysteine through transsulfuration (and then to taurine and/or the reduced form of glutathione, GSH) has been proposed as more effective in early pregnancy, in relation to the non-pregnant state26. This is indeed consistent with the observed early increase in excreted choline (Table 2, additional file 1, Figure S2) (possibly due to its reduced use in the remethylation pathway to produce DMG), as well as with the tendency for higher taurine levels in the 1st T (Supporting information, Figure S2). In later gestation, the same authors26 report the preferential activation of homocysteine



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 36

remethylation, which can occur in connection with the choline/DMG pathway or demethylation of tetrahydrofolate (THF). The increasing tendency seen for choline across the whole pregnancy period (Supporting information, Figure S2), although not statistically significant, suggests that the THF pathway may be favored, possibly in connection to folic acid diet supplementation (confirmed here for 76% of the pregnant women, information not available for the remaining). The concomitant lower taurine levels, in 2nd and 3rd trimesters (Table 2, Figure S2), are in agreement with the above mentioned redirection of the methionine pathway. In regard to the branched amino acids (BCAA) leucine and isoleucine, their variations differ from those reported previously28,29, both increasing significantly in the 3rd trimester. BCCA oxidation is known to decrease in late pregnancy thus increasing BCAA availability for the fetus1,20; similarly to other amino acids, this seems to lead to their enhanced excretion. STOCSY results showed that leucine and isoleucine integrals were consistently inter-correlated in all groups (r 0.68, 0.68, 0.63 and 0.85 in NP, 1st, 2nd and 3rd trimesters, respectively), reflecting their common pathway involvement. In addition to this, in the 1st T, isoleucine showed correlations to carnitine (r 0.72) and U4 (r 0.73), the latter two resonances also found to correlate to 3-HBA (r 0.74 and 0.93, respectively for carnitine and U4). This suggests a particular importance of isoleucine in lipid oxidation and ketone body synthesis, early on in pregnancy (even though increased maternal fat deposition and lipogenesis are known to take place preferentially, at that stage). Enhanced lipid oxidation later in gestation is consistent with the marked decrease of excreted carnitine in the 3rd T, confirming previous reports27, and with the accumulation of urinary 3-HBA. In the same trimester, this latter compound was found to correlate with 4-DEA and 4-DTA (r 0.73 and r 0.64, respectively), suggesting a possible relationship of these threonine


Page 15 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


catabolic products41,42 to ketone body production, consistently with the ketogenic nature of the amino acid. Regarding amino acid derivatives, changes were noted for the first time to our knowledge, in the excretion of glycine derivatives guanidoacetate, or GAA, and furoylglycine. High excreted levels of GAA have been previously associated to the accumulation of arginine in the urea cycle43, thus possibly relating to the expected slowing down of that pathway in pregnancy. This possible involvement in the urea cycle is consistent with a correlation found for GAA with alanine (r 0.61) in the 1st trimester (arginine has not been detected in this work). In relation to the variation in furoylglycine, a conjugation product of glycine with 2-furoic acid, (Table 2, Figure S2), it is well known that this compound relates strongly to diet, having been related to the consumption of chocolate and heated juices44. Its relation to diet was confirmed here by its almost 2-fold increase in the non-fasting test group (Supporting information, Figure S4b) so that it is possible, therefore, that the small increase (0.96 size effect) in furoylglycine in the 3rd T may arise from a spurious diet effect. However, the significant decrease in the 1st T (ca. 2-fold, p ≈ 10-5, Table 2) allows for the suggestion that a specific relationship of furoylglycine to pregnancy may exist, possibly in relation to lipid metabolism as is the case of other acyl-glycines8,45. Finally, still in relation to energy metabolism, it is interesting to note that the documented tendency for increased excretion of glucose in pregnancy28,29 was not detected as significant in the specific conditions of this study, probably indicating its reduced magnitude compared to the impact of effects

such as inter-subject variability and non-fasting. Furthermore, other changes found

previously in lactose28 were not observed in this study, whereas lower abundance sugars (ribose, xylose, fucose, fructose) and vitamins (nicotinic acid, folic acid), also expected to change, could not be detected.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 36

Creatinine metabolism Creatinine excretion was found to decrease significantly with pregnancy (Table 2, Figure S2), for the 1st time to our knowledge, in spite of earlier evidence that 24 hour urinary creatinine excretion was not altered during pregnancy46. Creatinine is formed from creatine, which is obtained either from diet or from the conversion of GAA, its levels in serum known to decrease throughout pregnancy due to increased glomerular filtration rate (GFR)29. The observed decrease in creatinine excretion may reflect the latter effect, however, other variables may contribute to it, namely a possible alteration of creatine synthesis in relation to urea cycle defects43 or the effect of diet47,48. In relation to the latter, a relative decrease in excreted creatinine was indeed observed in samples collected non-fasting (Supporting information, Figure S4), however, since all samples in our main study have been collected in such conditions, the consistent decrease across pregnancy may also reflect gestation itself.

Important unassigned spin systems Within the four unassigned resonances seen to change throughout pregnancy, U1 (0.55, s and 0.78 ppm, m) and U2 (0.63 ppm, s) stand out as highly significant. The possibility of these arising from bile acids was investigated through comparison with the spectra of standard compounds (namely, cholic acid, glycocholic acid, chenodeoxycholic acid, deoxycholic acid and glycochenodeoxycholic acid), however, no conclusive assignments were achieved, at this stage. Both resonances are positively correlated to each other, in all subject groups however, since their relative intensities are variable, it may be concluded that they originate from metabolically related entities rather than from one single compound. In the 3rd T, U2 further correlates to 3-


Page 17 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


HBA (r 0.74), which suggests a possible relation to ketone body formation. Regarding U3 (1.18 ppm, d; 1.45 and 3.82 ppm), correlations have been observed in 3rd trimester with leucine (r 0.64) and isoleucine (r 0.85), suggesting that the compound may be involved in the expected alterations of BCAA oxidation later in gestation. In relation to U4 (2.15 ppm, s), the noted positive correlations with 3-HBA and isoleucine (r 0.93 and r 0.73, respectively) also suggest a possible involvement in ketone body synthesis.

Conclusions This work employed a NMR metabonomics approach to study the variations in the composition of maternal urine of independent healthy subject groups, representing the non-pregnant state and each of the three pregnancy trimesters. The results obtained enabled 21 relevant metabolites to be identified and their variation throughout pregnancy to be followed, thus defining a 21metabolite signature for each pregnancy trimester of normal pregnancy. Within the variations noted, those in choline, creatinine, 4-DEA and 4-DTA, furoylglycine, GAA, 3-HBA and lactate (the latter with a residual change) were here observed for the first time, to our knowledge, in connection to pregnancy. The general increase in amino acid excretion confirmed the expected aminoaciduria accompanying pregnancy, with leucine and isoleucine being more significantly excreted than expected based on the literature, possibly in connection to a marked slowing down of BCAA oxidation in the 3rd T. With basis on STOCY analysis, isoleucine was also suggested to be somehow involved in lipid oxidation (through carnitine) and ketone body synthesis (through 3HBA), from early on in pregnancy. 3-HBA could be further associated to threonine and its degradation products, 4-DEA and 4-DTA, possibly in the context of ketogenesis. The newly



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 36

noted change in choline was suggested to relate, together with taurine, to methionine metabolism and preferential homocysteine methylation through THF, later in pregnancy. Furthermore, the over excretion of glycine derivative GAA was suggested to relate to the urea cycle deregulation known to accompany pregnancy, whereas another glycine derivative often associated with diet, furoylglycine, was proposed to have a possible particular relationship to pregnancy, due to its marked change in pregnant women, compared to the NP state. A similar proposal was advanced in relation to creatinine, in spite of its concomitant clear dependence on diet. Finally, it is important to note the marked dependence of certain unassigned resonances, particularly those named as U1 (0.55 ppm) and U2 (0.63 ppm), throughout pregnancy and which qualify as potential healthy pregnancy markers, albeit the presently hampered knowledge as to their exact origin. The results of this work show the usefulness of the untargeted compositional study of maternal urine, in order to achieve a multidirectional view of pregnancy metabolism and its adaptations as a function of time. This work enabled the definition of a metabolic dynamic signature for healthy pregnancies, unveiling some unexpected metabolic aspects and setting the basis for a control trajectory against which disease-related deviations may be confronted, in a clinical environment, serving as the basis for improved disease diagnostics and prediction.

Supporting Information Available: This material is available free of charge via the Internet at http://pubs.acs.org.

Acknowledgements


Page 19 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Funding is acknowledged from the European Regional Development Fund (FEDER) through the Competitive Factors Thematic Operational Programme (COMPETE) and from the Foundation for Science and Technology (FCT), Portugal, under the projects Pest-C/CTM/LA0011/2011, PTDC/QUI/66523/2006 and grant SFRH/BD/64159/2009. The authors are grateful to the Portuguese National NMR Network (RNRMN), supported with FCT funds, and to M. Spraul, Bruker BioSpin, Germany, for providing access to spectral databases.

Authors' contributions SD carried out the NMR analysis of urine samples, the spectral data analysis and drafted the manuscript. AB helped defining the multivariate analysis and validation protocols and aided in their interpretation. BJG and ID aided in the definition of the graphics and in the biochemical interpretation of results. IMC, EG, CP and MCA helped to define the design and setting up of sampling protocols in the clinic, subject guidance and information on the project, as well as in the biochemical interpretation of results. AMG conceived the study, participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.

References 1. Hod, M.; Renzo, G. C. D.; Leiva, A. D.; Langer, O. Ed. Textbook of diabetes and pregnancy; 2nd ed.; Informa Healthcare: London, U.K., 2008. 2. Hacker, N. F.; Moore, J. G.; Gambone, J. C. Ed. Essentials of obstetrics and gynecology; 4th ed.; Elsevier Saunders: Philadelphia, U.S., 2004.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 36

3. Tea, I.; Le Gall, G.; Kuster, A.; Guignard, N.; Alexandre-Gouabau, M-C.; Darmaun, D.; Robins, R. J. 1H-NMR-based metabolic profiling of maternal and umbilical cord blood indicates altered materno-foetal nutrient exchange in preterm infants. PLoS One 2012, 7, e29947. 4. Graça, G.; Goodfellow, B. J.; Barros, A. S.; Diaz, S.; Duarte, I. F.; Spagou, K.; Veselkov, K. ; Want, E. J.; Lindon, J. C.; Carreira, I. M.; Galhano, E.; Pita, C.; Gil, A. M. UPLC-MS metabolic profiling of second trimester amniotic fluid and maternal urine and comparison with NMR spectral profiling for the identification of pregnancy disorder biomarkers. Mol. BioSys. 2012, 8, 1243-1254. 5. Diaz, S. O.; Pinto, J.; Graça, G.; Duarte, I. F.; Barros, A. S.; Galhano, E.; Pita, C.; Almeida, M. C.; Goodfellow, B. J.; Carreira, I. M.; Gil, A. M. Metabolic biomarkers of prenatal disorders: an exploratory NMR metabonomics study of 2nd trimester maternal urine and blood plasma. J. Proteome Res. 2011, 10, 3732-3742. 6. Horgan, R. P.; Broadhurst, D. I.; Dunn, W. B.; Brown, M.; Heazell, A. E. P.; Kell, D. B.; Baker, P. N.; Kenny, L. C. Changes in the metabolic footprint of placental explantconditioned medium cultured in different oxygen tensions from placentas of small for gestational age and normal pregnancies. Placenta 2010, 31, 893-901. 7. Kenny, L. C.; Broadhurst, D. I.; Dunn, W.; Brown, M.; North, R. A.; McCowan, L.; Roberts, C.; Cooper, G. J. S.; Kell, D. B.; Baker, P. N. Robust early pregnancy prediction of later preeclampsia using metabolomic biomarkers. Hypertension 2010, 56, 741-749. 8. Dunn, W.; Brown, M.; Worton, S.; Davies, K.; Jones, R.; Kell, D.; Heazell, A., The metabolome of human placental tissue: investigation of first trimester tissue and changes related to preeclampsia in late pregnancy. Metabolomics 2012, 8, 579-597.


Page 21 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


9. Heazell, A. E. P.; Bernatavicius, G.; Warrander, L.; Brown, M. C.; Dunn, W. B., A Metabolomic Approach Identifies Differences in Maternal Serum in Third Trimester Pregnancies That End in Poor Perinatal Outcome. Reprod Sci 2012, 19, 863-875. 10. Lindon, J. C.; Holmes, E.; Bollard, M. E.; Stanley, E.G.; Nicholson, J.K. Metabonomics technologies and their applications in physiological monitoring, drug safety assessment and disease diagnosis. Biomarkers 2004, 9, 1-31. 11. Madsen, R.; Lundstedt, T.; Trygg, J. Chemometrics in metabolomics- A review in human disease diagnosis. Anal. Chim. Acta 2010, 659, 23-33. 12. Turner, E.; Brewster, J. A.; Simpson, N. A. B.; Walker, J. J.; Fisher, J. Plasma from women with preeclampsia has a low lipid and ketone body content— a nuclear magnetic resonance study. Hypertens. Pregnancy 2007, 26, 329-342. 13. Turner, E.; Brewster, J. A.; Simpson, N. A. B.; Walker, J. J.; Fisher, J. Aromatic amino acid biomarkers of preeclampsia - A nuclear magnetic resonance investigation. Hypertens. Pregnancy 2008, 27, 225-235. 14. Dunn, W. B.; Brown, M.; Worton, S. A.; Crocker, I. P.; Broadhurst, D.; Horgan, R.; Kenny, L. C.; Baker, P. N.; Kell, D.B.; Heazell, A. E. P. Changes in the metabolic footprint of placental explant-conditioned culture medium identifies metabolic disturbances related to hypoxia and pre-eclampsia. Placenta 2009, 30, 974-980. 15. Bock, J. L. Metabolic profiling of amniotic fluid by proton nuclear magnetic resonance spectroscopy: correlation with fetal maturation and other clinical variables. Clin. Chem. 1994, 40, 56-61. 16. Groenen, P. M. W.; Engelke, U. F.; Wevers, R. A.; Hendriks, J. C. M.; Eskes, T. K. A. B.; Merkus, H. M. W. M.; Steegers-Theunissen, R. P. M. High-resolution 1H NMR spectroscopy



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 36

of amniotic fluids from spina bifida fetuses and controls. Eur. J. Obstet. Gyn. R. B. 2004, 112, 16-23. 17. Graça, G.; Duarte, I. F.; Barros, A. S.; Goodfellow, B. J.; Diaz, S.; Carreira, I. M.; Couceiro, A. B.; Galhano, E.; Gil, A. M. 1H NMR based metabonomics of human amniotic fluid for the metabolic characterization of fetus malformations. J. Proteome Res. 2009, 8, 4144-4150. 18. Graça, G.; Duarte, I. F.; Barros, A. S.; Goodfellow, B. J.; Diaz, S. O.; Pinto, J.; Carreira, I. M.; Galhano, E.; Pita, C.; Gil, A. M. Impact of prenatal disorders on the metabolic profile of second trimester amniotic fluid: a nuclear magnetic resonance metabonomic study. J. Proteome Res. 2010, 9, 6016-6024. 19. Horgan, R. P.; Broadhurst, D. I.; Walsh, S. K.; Dunn, W. B.; Brown, M.; Roberts, C. T.; North, R. A.; McCowan, L. M.; Kell, D. B.; Baker, P. N.; Kenny, L. C. Metabolic profiling uncovers a phenotypic signature of small for gestational age in early pregnancy. J. Proteome Res. 2011, 10, 3660-3673. 20. King, J. C. Physiology of pregnancy and nutrient metabolism. Am. J. Clin. Nutr. 2000, 71, 1218S-1225S. 21. Herrera, E. Lipid metabolism in pregnancy and its consequences in the fetus and newborn. Endocrine 2002, 19, 43-55. 22. Sims, C. J.; Fujito, D. T.; Burholt, D. R.; Dadok, J.; Wilkinson, D. A. Comparison of metabolite levels in second and third trimester human amniotic fluid samples using proton magnetic resonance spectroscopy. J. Matern-Fetal Inves. 1996, 6, 62-66. 23. Cohn, B. R.; Fukuchi, E. Y.; Joe, B. N.; Swanson, M. G.; Kurhanewicz, J.; Yu, J.; Caughey, A. B. Calculation of gestational age in late second and third trimesters by ex vivo magnetic resonance spectroscopy of amniotic fluid. Am. J. Obstet. Gynecol. 2010, 203, 76.e1-76.e10.


Page 23 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


24. Ottolenghi, C.; Abermil, N.; Lescoat, A.; Aupetit, J.; Beaugendre, O.; Morichon-Delvallez, N.; Ricquier, D.; Chadefaux-Vekemans, B.; Rabier, D. Gestational age-related reference values for amniotic fluid organic acids. Prenatal Diag. 2010, 30, 43-48. 25. Athanasiadis, A. P.; Michaelidou, A-M.; Fotiou, M.; Menexes, G.; Theodoridis, T. D.; Ganidou, M.; Tzevelekis, B.; Assimakopoulos, E.; Tarlatzis, B. C. Correlation of 2nd trimester amniotic fluid amino acid profile with gestational age and estimated fetal weight. J. Matern.-Fetal Neo. M. 2011, 24, 1033-1038. 26. Dasarathy, J.; Gruca, L. L.; Bennett, C.; Parimi, P. S.; Duenas, C.; Marczewski, S.; Fierro, J. L.; Kalhan, S. C. Methionine metabolism in human pregnancy. Am. J. Clin. Nutr. 2010, 91, 357-365. 27. Cho, S-W.; Cha, Y-S. Pregnancy increases urinary loss of carnitine and reduces plasma carnitine in Korean women. Br. J. Nutr. 2005, 93, 685-691. 28. Hytten, F. E. The renal excretion of nutrients in pregnancy. Postgrad. Med. J. 1973, 49, 625629. 29. Creasy, R. K.; Resnik, R.; Iams, J. D. Ed. Creasy and Resnik's maternal-fetal medicine: principles and practice, 6th ed.; Saunders Elsevier: Philadelphia, U.S., 2009. 30. Wishart, D. S.; Knox, C.; Guo, A. C.; Eisner, R.; Young, N.; Gautam, B.; Hau, D. D.; Psychogios, N.; Dong, E.; Bouatra, S.; Mandal, R.; Sinelnikov, I.; Xia, J.; Jia, L.; Cruz, J. A.; Lim, E.; Sobsey, C. A.; Shrivastava, S.; Huang, P.; Liu, P.; Fang, L.; Peng, J.; Fradette, R.; Cheng, D.; Tzur, D.; Clements, M.; Lewis, A.; De Souza, A.; Zuniga, A.; Dawe, M.; Xiong, Y.; Clive, D.; Greiner, R.; Nazyrova, A.; Shaykhutdinov, R.; Li, L.; Vogel, H. J.; Forsythe, I. HMDB: a knowledgebase for the human metabolome. Nucleic Acids Research 2009, 37 (suppl 1), D603-D610.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 36

31. Veselkov, K. A.; Lindon, J. C.; Ebbels, T. M. D.; Crockford, D.; Volynkin, V. V.; Holmes, E.; Davies, D. B.; Nicholson, J. K. Recursive segment-wise peak alignment of biological 1H NMR spectra for improved metabolic biomarker recovery. Anal. Chem. 2009, 81, 56-66. 32. Dieterle, F.; Ross, A.; Schlotterbeck, G.; Senn, H. Probabilistic quotient normalization as robust method to account for dilution of complex biological mixtures. Application in 1H NMR metabonomics. Anal. Chem. 2006, 78, 4281-4290. 33. Joliffe, I. T.; Morgan, B. J. Principal component analysis and exploratory factor analysis. Stat. Methods Med. Res. 1992, 1, 69-95. 34. Barker, M.; Rayens, W.; Partial least squares for discrimination. J. Chemometr. 2003, 17, 166-173. 35. Wiklund, S.; Nilsson, D.; Eriksson, L.; Sjöström, M.; Wold, S.; Faber, K. A randomization test for PLS component selection. J. Chemometr. 2007, 21, 427-439. 36. Westerhuis, J.; Hoefsloot, H.; Smit, S.; Vis, D.; Smilde, A.; van Velzen, E.; van Duijnhoven, J.; van Dorsten, F. Assessment of PLSDA cross validation. Metabolomics 2008, 4, 81-89. 37. Nakagawa, S.; Cuthill, I. C., Effect size, confidence interval and statistical significance: a practical guide for biologists. Biol Rev 2007, 82, 591-605. 38. Cloarec, O.; Dumas, M. E.; Craig, A.; Barton, R. H.; Trygg, J.; Hudson, J.; Blancher, C.; Gauguier, D.; Lindon, J. C.; Holmes, E.; Nicholson, J. Statistical total correlation spectroscopy: an exploratory approach for latent biomarker identification from metabolic 1H NMR data sets. Anal. Chem. 2005, 77, 1282-1289. 39. Llorach, R.; Garcia-Aloy, M.; Tulipani, S.; Vazquez-Fresno, R.; Andres-Lacueva, C., Nutrimetabolomic strategies to develop new biomarkers of intake and health effects. J. Agric. Food Chem. 2012, 60, 8797-8808.


Page 25 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


40. Miller, S.; Ruttinger, V.; Macy, I. G. Urinary excretion of ten amino acids by women during the reproductive cycle. J. Biol. Chem. 1954, 209, 795-801. 41. Appiah-Amponsah, E.; Shanaiah, N.; Nagana Gowda, G. A.; Owusu-Sarfo, K.; Ye, T.; Raftery, D. Identification of 4-deoxythreonic acid present in human urine using HPLC and NMR techniques. J. Pharmaceut. Biomed. 2009, 50, 878-885. 42. Ellis, J.; Athersuch, T.; Thomas, L.; Teichert, F.; Perez-Trujillo, M.; Svendsen, C.; Spurgeon, D.; Singh, R.; Jarup, L.; Bundy, J.; Keun, H. C. Metabolic profiling detects early effects of environmental and lifestyle exposure to cadmium in a human population. BMC Med. 2012, 10, 61-71. 43. Brosnan, J. T.; Brosnan, M. E. Creatine metabolism and the urea cycle. Mol. Genet. Metab. 2010, 100 (Suppl 1), S49-S52. 44. Kumps, A.; Duez, P.; Mardens, Y. Metabolic, nutritional, iatrogenic, and artifactual sources of urinary organic acids: a comprehensive table. Clin. Chem. 2002, 48, 708-717. 45. Ombrone, D.; Salvatore, F.; Ruoppolo, M., Quantitative liquid chromatography coupled with tandem mass spectrometry analysis of urinary acylglycines: Application to the diagnosis of inborn errors of metabolism. Anal. Biochem. 2011, 417, 122-128. 46. Gallery, E. D. M.; Ross, M.; Gyory, A. Z., 24-Hour urinary creatinine excretion is not altered in human pregnancy. Hypertens. Pregnancy 1996, 15, 257-261. 47. Solanky, K. S.; Bailey, N. J.; Beckwith-Hall, B. M.; Bingham, S.; Davis, A.; Holmes, E.; Nicholson, J. K.; Cassidy, A. Biofluid 1H NMR-based metabonomic techniques in nutrition research metabolic effects of dietary isoflavones in humans. J. Nutr. Biochem. 2005, 16, 236244.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 36

48. Stella, C.; Beckwith-Hall, B.; Cloarec, O.; Holmes, E.; Lindon, J. C.; Powell, J.; van der Ouderaa, F.; Bingham, S.; Cross, A. J.; Nicholson, J. K. Susceptibility of human metabolic phenotypes to dietary modulation. J. Proteome Res. 2006, 5, 2780-2788. 49. Maher, A. D.; Cysique, L. A.; Brew, B. J.; Rae, C. D. Statistical Integration of 1H NMR and MRS Data from Different Biofluids and Tissues Enhances Recovery of Biological Information from Individuals with HIV-1 infection. J. Proteome Res. 2011, 10, 1737-1745. 50. Rivière, C.; Thi Hong, V. N.; Hoai, N. N.; Dejaegher, B.; Tistaert, C.; Van, K. P.; Heyden, Y. V.; Chau Van, M.; Quetin-Leclercq, J. N-methyl-5-carboxamide-2-pyridone from Mallotus barbatus: A chemosystematic marker of the Euphorbiaceae genus Mallotus. Biochem. Syst. Ecol. 2012, 44, 212-215. 51. Rezzi, S.; Ramadan, Z.; Martin, F. P. J.; Fay, L. B.; van Bladeren, P.; Lindon, J. C.; Nicholson, J. K.; Kochhar, S. Human metabolic phenotypes link directly to specific dietary preferences in healthy individuals. J. Proteome Res. 2007, 6, 4469-4477.


Page 27 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figures and Tables Captions Figure 1. 500MHz 1H-NMR urine spectra of a) non-pregnant and pregnant women at b) 12, c) 17, d) 35 g.w.. Assignment (with 3-letter codes for a.a.): 1) Ile 2) Leu, 3) Val, 4) 3hydroxybutyrate, 3-HBA, 5) 3-aminoisobutyrate, 6) 3-hydroxyisovalerate, 7) lactate, 8) Thr, 9) 2-hydroxyisobutyrate, 10) Ala, 11) Lys, 12) acetate, 13) phenylacetylglutamine, 14) p-cresol sulphate, 15) succinate, 16) citrate, 17) dimethylamine, 18) creatine, Cr, 19) creatinine, Crn, 20) malonate, 21) cis-aconitate, 22) choline, 23) carnitine, 24) betaine, 25) trimethylamine-N-oxide, TMAO, 26) taurine, Tau, 27) Gly, 28) guanidoacetate, GAA, 29) trigonelline, 30) glucose, 31) urea, 32) furoylglycine, 33) N-methyl-2-pyridone-5-carboxamide, 2PY, 34) Tyr, 35) His, 36) hippurate, 37) hypoxanthine, 38) formate. The rectangles guide the eye for some visible spectral changes. Remaining assignments are shown in supporting information, Table S2.

Figure 2. PLS-DA score (left) and loading plots (right) for a) NP(□) vs 1stT(♦), b) 1stT(♦) vs 2ndT(○) and c) 2ndT(○) vs 3rdT(▲)groups. Unassigned spin systems (Uni) are numbered consistently with Table 2. Loadings are colored according to Variable Importance to the Projection (VIP). Three-letter codes are used for amino acids; GAA: guanidoacetate; 4-DTA: 4deoxyerythronic acid; 4-DTA: 4-deoxythreonic acid. The circles indicate two outlier samples, one of the 1st trimester and the other of the 3rd.

Figure 3. Heatmap of normalized integrals of the 21 metabolites varying across pregnancy. Lines and columns represent subjects and metabolites, respectively. Integrals are shown in a color scale from minimum (dark blue) to maximum (dark red) values. Metabolites are ordered from those exhibiting marked increases (left) to those showing significant decreases (right).



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 36

Metabolites observed to change in maternal in urine, throughout pregnancy, for the first time to our knowledge.

Figure 4. PCA and PLS-DA scores for 1H-NMR urine spectra of NP(□), 1stT(♦), 2ndT(○), 3rdT(▲) pregnant women. a) PCA and b) PLS-DA scores plots obtained using all spectral points and c) PCA and d) PLS-DA scores plots obtained using the 21 most relevant integrals (p-value < 0.05). The circles in a) and b) indicate the 1st T outlier shown in Figure 2 (the 3rd T outlier seen in Figure 2 is not visible in these plots). The curved arrow in d) indicates the trajectory followed across pregnancy.


Page 29 of 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


TABLES

Table 1. List of urine samples collected for each independent group of subjects, along with corresponding ranges of maternal age (in years), pre-pregnancy body mass index (BMI, in kg.m2

), gestational age at sampling and at delivery (in gestational weeks), gravidity (no. pregnancies),

parity (no. born children), newborn’s birth weight (grams), no. pregnant women who were smokers and who were taking folic acid supplementation at the time of sampling. Median values are shown in brackets.

NP

1st Trimester (1st T)

2nd trimester (2nd T)

3rd trimester (3rd T)

16

16

20

19

Maternal age range

21-47 (26)

19-40 (31)

21-39 (34)

20-38 (32)

BMI range

18-35 (22)

19-25 (24)

20-33 (24)

17-35 (23)

Gestational age at time of sampling

-

11-13 (12)

15-26 (17)

29-39 (35)

Gestational age at delivery

-

37-40 (40)

37-40 (38)

37-40 (38)

Gravidity

-

1-2 (1)

1-6 (2)

1-5 (2)

Parity

-

0-1 (0)

0-2 (1)

0-3 (0)

Newborns’ birth weight

-

2590-4110 (3398)

2610-3740 (3238)

2330-3950 (3274)

No. smokers

-

0

1

2

No. subjects taking folic acid at time of sampling

-

12

13

17

n



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 36

Table 2. List of metabolite/resonance changes in urine from NP state to 1st T, from 1st to 2nd T and from 2nd to 3rd T. a Chemical shifts shown correspond to signals used for integration; values in square brackets correspond to correlated resonances seen through either TOCSY or STOCSY; s: singlet, d: doublet, t: triplet, q: quartet, dd: doublet of doublets, m: multiplet. b Significance level: 95% (p-value

Metabolic Profiling of Human Urine - American Chemical Society

Recommend Documents